- From: Daniel LaLiberte <liberte@ncsa.uiuc.edu>
- Date: Fri, 21 Apr 95 09:44:58 CDT
- To: Larry Masinter <masinter@parc.xerox.com>
- Cc: uri@bunyip.com
Larry Masinter writes: > Minutes of Uniform Resources Identifiers Working Group, April 3 & 4, > 1995. > * Michael Shapiro presented the Path URN scheme. > - Totally separate from hostnames, but have to emulate what DNS > is doing now. This is potentially confusing. The idea is that if the path scheme does not use DNS for its implementation, we would have to recreate the features of DNS used by the path scheme. > 6. The remainder of the session was given over to continued discussion > of the 5 URL schemes and associated issues. > > * Most of the URN schemes don't really go into details about what > happens when something moves. So far have dealt only with what > happens when a whole organization moves. Granularity of mobility must > be addressed. The description of the path scheme should have made it clear that the path scheme does allow fine grain mobility. There are two mechanisms to support this, one for directories and one for individual documents. If a subdirectory of documents is moved to a new server, the resolution process for looking up a name of one of those documents will find the new server because it is more specific. (Name servers near the client can cache the location of the new server.) If just a single document is moved, the original server would forward references to the current location. > For example: > > - All web documents from CERN move to INRIA, but high-energy physics > documents stay at CERN. What happens to URNs for the web documents? Depending on how the documents are layed out (how they are named) this would probably be handled by the first mechanism that supports subdirectory relocation. > - A grad-student graduates but someone wants to retain editorial > control of one of his pages. How can this happen? If the someone who wants to retain editorial control of the page is associated with the original server, then the student should agree to forward requests of the page back to the original server in exchange for the original server agreeing to forward requests for all the student's documents to the new server. Perhaps the original server should never have given the student control of such a document in the first place if it planned to retain control. Perhaps a flat name system could handle these change of control issues better. The path scheme could have a mechanism to essentially appeal to a higher authority, in case the immediate parent authority doesn't want to cooperate. Perhaps the path scheme already allows this appeal process - have to think about it some more. This is where the hierarchical path scheme starts to merge with the centralized handle scheme. > - A university retains all articles by staff members; someone on the > staff publishes the exact document in an online journal. Does it > get a new URN, or just keep the old one? This is a question not about mobility but about the abstraction of names of things that are replicated. Each of the university copy and the online journal copy could have individual URNs, and another URN could be assigned to the abstraction over both. Caches should use the more abstract URN, unless there is something to distinguish each of the copies, such as the online journal copy costs more. > * Are we trying to solve NIDR with permanent names? Names should last > as long as the issuing agency... go to a permanent naming authority > if you want your stuff to last. Or the resolution of new names could be controlled by issuing agency as in the path scheme, but the resolution of old but still valuable names could migrate up to a permanent central resolution service. > * Some assert that it's time to decide on a syntax and character set > so experimental work can go forward. On the other hand, we must agree > on the semantics before we do a syntax. There is no point in considering a common syntax across all URNs (other than the generic URL syntax) unless it makes sense to have a common semantics across all URNs. I don't believe it makes sense to think that we can come up with a common semantics for all possible name schemes, now existing or created in the future. The generic URL syntax allows any semantics, and that is what we need in general. Daniel LaLiberte (liberte@ncsa.uiuc.edu) National Center for Supercomputing Applications http://union.ncsa.uiuc.edu/~liberte/
Received on Friday, 21 April 1995 10:48:56 UTC